Origin and Scope of Long-Range Stabilizing Interactions and Associated SOMO − HOMO Conversion in Distonic Radical Anions

Ganna Gryn’ova and Michelle L. Coote*

ARC Centre of Excellence for Free Radical Chemistry and Biotechnology, Research School of Chemistry, Australian National University, Canberra, Australian Capital Territory 0200, Australia

*S Supporting Information

ABSTRACT: High-level quantum-chemical methods have been used to study the scope and physical origin of the signifi_{cant long-range stabilizing interactions between non-} mutually conjugated anion and radical moieties in SOMO−

HOMO converted distonic radical anions. In such species, deprotonation of the acid fragment can stabilize the remote radical by tens of kilojoules, or, analogously, formation of a stable radical (by abstraction or homolytic cleavage reactions) increases the acidity of a remote acid by several pKaunits. This

stabilization can be broadly classifi_{ed as a new type of polar}

eff_{ect that originates in Coloumbic interactions but, in contrast to standard polar e}ff_{ects, persists in radicals with no charge-} separated (i.e., dipole) resonance contributors, is nondirectional, and hence of extremely broad scope. The stabilization upon deprotonation is largest when a highly delocalized radical is combined with an initially less stable anion (i.e., the conjugate base of a weaker acid), and is negligible for highly localized radicals and/or stable anions. The eff_{ect is largest in the gas phase and low-} polarity solvents but is quenched in water, where the anion is suffi_{ciently stabilized. These simple rules can be employed to} design various switchable compounds able to reversibly release radicals in response to pH for use in, for example, organic synthesis or nitroxide-mediated polymerization. Moreover, given its wide chemical scope, this eff_{ect is likely to in}fl_{uence the} protonation state of many biological substrates under radical attack and may contribute to enzyme catalysis.

■

INTRODUCTION

Radical ions are fascinating species that possess dual reactivity, and both modesradical and ioniccan greatly aff_{ect one} another. They are involved in many crucial biological processes and play an increasingly important role in a range of practical applications.1,2_{Radical ions can be broadly divided into two}

families: classic (conventional) in which the charge and unpaired spin are located on the same atom or conjugated fragment (e.g., as formed by electron addition to or removal from a neutral closed-shell species) and distonic, in which charge and spin are spatially separated and which are formed by ionization of zwitterions or diradicals.3 _{The former are}

interesting in their own right and have been extensively studied in relation to their role in the radiation-induced damage of nucleic acids4_{and as a convenient tool for measuring gas- and}

solution-phase thermochemistry.5 _{Seemingly exotic, distonic}

radical ions are actually quite common, often more stable than their conventional isomers and highly relevant to various research fi_{elds ranging from mass spectroscopy}6 _{to the} decomposition of drinking water pollutants, and the photo- chemical damage of amino acids and peptides.7

There is a growing appreciation of the often dramatic diff_{erences in the stability and reactivity of distonic radical ions} and their conventional radical ion, neutral radical, or charged closed-shell counterparts.8_{These differences can be employed,}

for example, to preferentially stabilize the classic or distonic

form, as illustrated by Pius and Chandrasekhar9 _{for the}

persistent organometallic radical anions •CH2−XH−. More generally, standard through-bond and through-space polar eff_{ects can alter not only the stability of the radical but also, by} extension, the strength of its bonds or the kinetics of its reactions.10 _{Kenttämaa et al.}11 _{showed that reactivity of a}

phenyl radical is signifi_{cantly altered in its distonic cation and} anion derivatives. Boyd et al.12 _{and later Radom and co-}

workers13_{found that, in general, protonation of X in CH3−X}

(where X = OH, NH2, Hal, CN, NO2, etc.) shortens and

strengthens the bond (against homolytic cleavage) and destabilizes the resulting carbon-centered radical, while deprotonation of X has an opposite eff_{ect, typically smaller in} magnitude. These properties have been explained in terms of orbital interactions, resonance stabilization and competing heterolytic dissociation. Similarly, N-protonation and N- deprotonation infl_{uences N}−_{H bond dissociation energy} (BDE) in carbamates.14 _{Finally, interactions between charge}

and spin act both ways, i.e. an unpaired electron can aff_{ect the} properties of the charged moiety, such as the strength of its conjugated acid. Indeed, abnormally low pKa values were

predicted for the carbon acidity of formic and acetic acid radicals RCOO•.15

Received: April 30, 2013 Published: September 18, 2013

Article

pubs.acs.org/JACS

However, such eff_{ects of a charge on a radical (and of a}

radical on a charge) are not unexpected, considering that in the above studies the two moieties are in ultimate close proximity, i.e. separated by only one or two chemical bonds. Such interactions can also exist at larger separationsas much as 10 Å and aboveprovided, however, that the radical and charge are directlyπ-conjugated, i.e. polar eff_{ects utilizing resonance}

(orbital overlap). For example, stability of peptide C-centered radicals can be aff_{ected noticeably by the protonation at the}

distal amide nitrogen.17 _{Another striking example is the}

weakening (by over 60 kJ mol−1_{) of one RCC−H bond in}

2,6-diethynylnaphtalene toward the homolytic cleavage due to the deprotonation of the other acetylene group.18Furthermore, Ingold et al. discovered surprisingly large effects (∼10 kJ mol−1

in thermodynamics and over 20 times speed-up in kinetics) of remote hydrogen bonds on the stability and reactivity of benzyl cations and phenoxyl radicals, arising in their charge-separated resonance contributors.19

In the absence of π-conjugation and, even more generally, any direct chemical bonding between the charge and radical, through-space polar eff_{ects can occur,}10a_{albeit they typically}

remain signifi_{cant within only a short-range of}∼4−5 Å. Such

eff_{ects are thought to arise in the Coloumbic interactions}

between the anion or cation and a permanent dipole, associated with the radical moiety, and hence are directional. For instance, C−H bonds can be selectively activated or deactivated toward homolytic cleavage by a remote amine radical cation in n- butylamine.20_{Furthermore, polar and inductivefield effects in}

transition states of H-abstraction from the peptide backbone and adjacent side-chain carbons by Cl• contribute to the resistance of these units against radical damage.21_{Finally, the}

destabilizing eff_{ect of a positive charge on a polar resonance}

contributor R1_R2_N•+_−O−_{of an aminoxyl radical R}1_R2_N−O•

(more commonly referred to as‘nitroxide’)22_{was employed to}

design a pH-switchable agent for nitroxide mediated polymer- ization (NMP).23

However, very recently we have discovered a signifi_cant

stabilizing interaction occurring between a truly remote negative charge and stable radical (in the absence of any π- or σ-conjugation or hyperconjugation and at long-range separations of over 5 Å, Figure 1),24 _{which are not}

straightforward to rationalize in terms of conventional physical organic chemistry concepts.25 _{Specifically, we found that}

deprotonation of an acid−base group (carboxylate, alkoxide, sulfate) results in an unprecedented stabilization of a remote radical, manifested in ∼20 kJ mol−1 _{decrease in dissociation}

energy (or 4 orders of magnitude in the corresponding logKeq)

of its bonds with carbon-centered radicals, as compared with the protonated or nonsubstituted forms. These computational results were verifi_{ed by comparison with gas-phase thermo-}

chemistry measurements, as obtained via mass spectrometry, and the mean absolute deviation between the computationally derived BDE-switches and their experimental counterparts was just 1.7 kJ mol−1_.24

We also showed that this unexpected stabilization decays linearly with 1/r(whereris the distance between formal charge and formal radical, see Figure 2), does not appear to require

chemical bonding between the two moieties (i.e., is a through- space rather than a through-bond eff_{ect) and is not associated}

with orbital overlap or net electron transfer.26At the same time, we found that this long-range pH eff_{ect on radical stabilization}

is only signifi_{cant if the corresponding neutral radical is}

relatively stable (delocalized) initially (aminoxyl R1_R2_NO•,

aminyl R1_R2_{N•, peroxyl ROO•). In less stable (more localized)}

radicals, such as alkoxyl RO•, the magnitude of such pH- switches on radical stability, which we defi_{ne as the BDE-switch}

= BDE[HA−X−R]−BDE[−_{A−X−R], is much smaller (less}

than 10 kJ mol−1_{). Admittedly, all of these species have dipole}

Figure 1.Properties of the 4-carboxylate-2,2′,6,6′-tetramethylpiperidine-N-oxyl (4-COO−_{-TEMPO, in bold) distonic radical anion, a thermocycle} relating switches on the radical BDE and the anion’s conjugate acid gas-phase acidity (GPA) switch, as well as energy breakdown of the switches in different compounds with similar separation between formal charge and formal radical (in italics). Curly arrows indicate investigated breaking bonds. All BDE-switches (kJ mol−1_{) were calculated using the G3(MP2,CC)(+)}16_method.

Figure 2. Energy breakdowns of the BDE-switches (y-axis, in kJ mol−1_{) plotted vs inverse separation rbetween formal charge and}

formal radical (x-axis, in Å−1_{); diameters of the ‘bubbles’ are}

proportional to the contributions of diff_{erent components to the net} BDE-switch. Negative contributions in the cyclic alkyl series are shown in white.

Journal of the American Chemical Society Article

dx.doi.org/10.1021/ja404279f|J. Am. Chem. Soc.2013, 135, 15392−15403 15393

contributors to their resonance stabilization, as shown for aminoxyl in Figure 1, and an appropriately placed charge of the correct sign would stabilize such a contributor and thus the radical overall. In addition, polar eff_{ects can in principle act on} the breaking bonds in the parent closed-shell compounds as well; e.g. a negative charge in the TEMPO ring is likely to destabilize a polar resonance contributor R1_R2_NO−_···+_{R of an}

alkoxyamine R1_R2_{NO−R, hence further decreasing the}

corresponding BDE.27 _{However, when we decomposed the}

net BDE-switches into ‘Correlation’ (diff_{erence between net} and Hartree−Fock (HF) energy),‘Exchange’(sum of exactα

andβexchange contributions to HF) and‘Hartree’(diff_erence between HF and Exchange) components, we found that these traditional directional polar eff_{ects arise mainly in the Exchange} component of the pH switch. As seen in Figure 2, this component constitutes almost the entire switch in localized RO• radicals and contributes to the ROO• and R1_R2_NO•

series. However, these latter24systems additionally involve an even greater Hartree contribution.28_{Furthermore, the corre-}

sponding alkyl series R•, which lack both the resonance stabilization and the dipole contributors in either form (open- or closed-shell), display negligible pH switching (Figure 2).

Intriguingly, we also showed that this unparalleled eff_{ect on} stability and reactivity of distonic radical anions is associated with an electronic structure phenomenon known as orbital conversion, in which the singly occupied molecular orbital (SOMO, corresponds to the unpaired electron) is not the highest one (HOMO) because one or more doubly occupied orbitals corresponding to anion have greater energies (see Figure S1 in the Supporting Information [SI]).24 This is surprising because normally in radicals the SOMO is the HOMO according to the aufbau principle, and such regular orbital occupation is indeed restored upon protonation of the anion. Yet, this behavior is not unprecedented, and has been observed in a very limited number of stable neutral radicals and developed into fascinating molecular electronics applications utilizing oxidation of such ‘converted’radicals into high-spin states.29_{Indeed, our orbital-converted distonic radical anions}

were shown to oxidize preferentially to triplets, whereas the corresponding protonated species yield even-electron prod- ucts.24_{Hence, stable distonic radical anions represent a new}

class of orbital converted compounds, and have the added advantage that the orbital conversion can be switched via pH. This existence of long-range stabilizing interactions between certain radicals and anions is likely to be useful in a broad range of practical applications, from reversibly pH-switchable radical protecting groups in organic synthesis, controlled radical polymerization and polymer end-group modifi_{cation, that can} release or trap radicals in response to pH changes, to reversible pH-switchable orbital conversion and associated oxidation to high-spin states for use in molecular electronics and sensing applications. At the same time, our original report of this unusual and useful eff_{ect admittedly raises more questions than} it answers.24

•What is the cause of the effect?Is orbital energy-level conversion the primary cause of the unexpected radical stabilization (and increased strength of anion’s con- jugated acid), or is it an accompanying feature of the unusually stable distonic radical anions? And if so, can the radical stabilization occur even in the absence of orbital conversion, i.e. caused by a negative point charge or electricfi_{eld? If the stabilizing e}ff_{ect is polar in nature,}

can positive charges as well as the negative ones generate it, and can it appear in species without charge-separated (dipole) contributors and hence act nondirectionally?

• What is the scope of the effect? Is it limited to combinations of only those few anions and radicals that we have already considered,24_{or is it more common and}

can occur in, for example, carbon-centered radicals? Does it require radicals with dipoles in their resonance contributors? Can the magnitude of the stabilizing interaction be manipulated by the stabilities of radical and anion?

• Is this new effect influenced by the external conditions, such as temperature or binding of anion to cations other than a proton, and is it preserved in the condensed phase?

Answers to these and associated questions are crucial for better understanding of the origin of this new type of long- range interaction between a charge and an unpaired spin, and the dramatic eff_{ect it has on both the chemical stability and} reactivity of the two moieties and the orbital confi_{guration of} the molecule comprising them. Moreover, they shape the practical applications of this discovery and its implications for various biological processes. In this work we use theory to address these important questions and explore the origin and the broad scope of the switching eff_{ect on radical and anion} stability and the associated orbital reordering.

■

COMPUTATIONAL METHODOLOGY

In order to explore the origin and scope of the long-range stabilizing interaction between the negative charge and the unpaired spin in

distonic radical anions, and their associated SOMO−HOMO energy-

level conversion, we have employed quantum-chemical methods of

varying computational cost and accuracy, includingab initio, density

functional theory (DFT) and several high-level composite Gn

methods. Our benchmark computational methodologies have been extensively tested against experimental data and shown to deliver

results to within the chemical accuracy (∼5 kJ mol−1 _{for bond}

dissociation energies and gas-phase acidities, and 0.050 V for redox

potentials).24,30−32 _{Moreover, both our previous results}24 _and

benchmarking for a representative test set in this work also reveal

generally good agreement between theswitchvalues calculated using

our benchmark methods and various lower-cost procedures. In many

cases this is due to fortuitous cancellation of errors, which is reflected

in less cohesive results for the calculated absolute BDEs and GPAs.

Nonetheless, of the lower-cost methods examined, M06-2X33_/6-

31+G(d) consistently exhibited excellent performance against the

available experimental data and high-level Gnresults for the switches

(mean absolute deviation from G3 is only 0.4 kJ mol−1_{). For}

consistency all switch values shown and discussed below are calculated with this method, however representative results obtained using more sophisticated methodologies are provided in the SI. Where the

absolute BDE and GPA values are involved, we either confirm M06-2X

accuracy via benchmarking against high-level composite theoretical methods and, where available, experimental values, or employ the

high-level G3(MP2,CC)(+) method16 _{instead. Our earlier}24 _and

present results indicate that there is no appreciable diff_{erence between}

the switch values, calculated from electronic energies and corrected for

thermal and entropic eff_{ects; thus, throughout this study we present}

the electronic energy switches, with the exception of the sections in which the thermochemical factors and solvation are discussed. All

calculations were performed usingGaussian09,34_QChem3.235_and

Molpro2009.136_{software packages. A complete set of the obtained}

results and details of all the theoretical procedures, including extensive benchmarking, can be found in the SI.

Journal of the American Chemical Society Article

dx.doi.org/10.1021/ja404279f|J. Am. Chem. Soc.2013, 135, 15392−15403

■

RESULTS AND DISCUSSION

We have employed rigorously benchmarked theoretical methodology to clarify the origin and outline the broad scope of the mutually stabilizing eff_{ect of remote anion and radical,}

accompanied by an orbital energy-level conversion in the corresponding distonic radical anions. Specifi_{cally, we have}

calculated absolute bond dissociation energies (BDEs) and gas- phase acidities (GPAs) of a large array of molecules and molecular complexes comprising chemically diverse acid and radical moieties (coupled with diff_{erent leaving groups) under}

various conditions, and ascertained the pH‘switches’on BDEs (from the diff_{erences between their protonated and deproto-}

nated forms) and GPAs (odd- and even-electron forms). We have also studied the eff_{ects of detached positive and negative}

point charges on these diff_{erent radicals. Using these results,}

which are outlined below, wefi_{rst determine the contributions}

of orbital conversion and polar interactions to the stabilizing eff_{ect, and then quantify and analyze the roles of radical}

stability, anion stability, and external conditions on its magnitude.

Role of SOMO−HOMO Orbital Conversion. To date, using quantum-chemical methods ranging from single-determi- nant DFT to multireference MRPT2 and CASSCF we have shown that deprotonation of a remote acidic group in certain stabilized radicals leads to unprecedented radical stabilization and SOMO−HOMO conversion; in cases where much more muted stabilization is observed, there is no associated orbital conversion.24 To assess whether SOMO−HOMO orbital conversion is the primary cause of the stabilization or merely associated with it, we have now considered a series of neutral aminoxyls containing nonconjugated aromatic heterocyclic fragments that also display SOMO−HOMO orbital conversion (as shown by quantum-chemical calculations in Figure S3 and Table S3 of the SI and, for selected compounds, evident from their experimentally observed oxidation to biradicals29a,b,e_{) but}

lack the remote negative charge. Additionally, we considered carbenes, both singlet and triplet, as an alternative neutral source of high-energy HOMO(s). For all of these neutral compounds, BDE-switches were assessed by comparison with the structurally similar but non-SOMO−HOMO converted NN (nitronyl nitroxide), TEMPO (2,2′,6,6′-tetramethylpiper- idine-N-oxyl), or TMAO (1,1′,3,3′-tetramethyl-azaphenalene-

In document Understanding and Manipulating The Reactivity of Nitroxides and Other Stable Free Radicals (Page 159-171)